People are speaking with ChatGPT for hours, bringing 2013’s Her closer to reality::Long mobile conversations with the AI assistant using AirPods echo the sci-fi film.

  • 1bluepixel@lemmy.world
    link
    fedilink
    English
    arrow-up
    37
    arrow-down
    1
    ·
    1 year ago

    I tried having a conversation with ChatGPT. It’s annoyingly predictable. Imagine the most boring, chronically helpful therapist who is always brimming with obvious advice, and that’s what you get.

    I get that people are lonely, but we’re still much closer to ELIZA than Her.

    • ISometimesAdmin@the.coolest.zone
      link
      fedilink
      arrow-up
      12
      ·
      1 year ago

      I think said lonely people would rather chat with a boring therapist than with assholes on the internet who don’t even make the slightest attempt to empathize

    • L_Acacia@lemmy.one
      link
      fedilink
      English
      arrow-up
      9
      ·
      edit-2
      1 year ago

      Llama models tuned for conversation are pretty good at it. ChatGPT also was before getting nerfed a million time.

      • bassomitron@lemmy.world
        link
        fedilink
        English
        arrow-up
        3
        ·
        1 year ago

        Yeah, I think a lot of people don’t realize that they’ve neutered the shit out of ChatGPT. They don’t want it to be vividly humanlike, just mostly so. The main corporate interest in chat AI is profit, i.e. how can they eliminate various repetitive jobs that haven’t been automated yet because they require someone with pseudo-humanlike behavior. Examples would include stuff like help desks, customer services, knowledge managers, certain types of assistants, legal aids, etc. That’s their end game.

        Remember, many of these companies diving into this field are very wary of their tech coming off as too realistic. Some of us may be excited at the prospect of AGI becoming a reality, but I would bet the majority of society (and likely many governments) would instantly turn on such tech. Even though LLMs are far, far, faaaar away from achieving AGI, the fact that they already freak many folks out with their current limitations proves my point.

        Anyway, sorry for the rant. TLDR, OpenAI intentionally keeps ChatGPT’s conversational abilities relatively simple/efficient since they’re focused more on it seeming human enough to get the job(s) done.

    • scarabic@lemmy.world
      link
      fedilink
      English
      arrow-up
      5
      ·
      1 year ago

      What you described is quite a bit better than what some people get from their lives. Sad but true.